[SOLVED] DC current source problem

[Rumpelstiltskin]

Hey!
I got this circuit to generate 0.3A DC current through load (below).

Load can vary as well as Vsupply. By varying Vsupply I found that minimum Vce of NPN pair for this circuit to be working is 1.6V (V_collector - V_emitter). However I find this value to be very high.

For the BC847 transistor, worst case values from the datasheet are:

• Vbe(sat) = 0.72V
  
• Vce(sat) = 0.25V
  
• hFE = 25 (depending on the specific variant)
  

For the BD139 transistor, worst case values from the datasheet are:

• Vbe(sat) = 1V
  
• Vce(sat) = 0.5V
  
• hFE = 40 (depending on the specific variant).
  

Voltage at Rset is 0.75V, worst case Vbe of the pair roughly 1.72V. Vb of the pair would be 0.75V + 1.72V = 2.47V. Current needed on the base is 0.3A / 25 / 40 = 0.3 mA. So opamp should do the job.
Minimum Vce of the NPN pair I calculate as follows: Vbe(sat) BD139 + Vce(sat) BC847 = 1.25V in order circuit to work.

Measured value is 1.6V minimum and I fail to see why. Is my estimation wrong or something's wrong with PCB? Maybe there's some sort of circuit instability takes place? (I don't have oscilloscope to verify last)

Components R18, R19, C19 are added for stability.

[Benta]

1.6 V is actually a good value, I'd expect around 2 V.
VCEsat is for hard switching, not for linear operation.
The absolute minimum you could expect is BD139 VBE + BC547 VCE, in your case 1 V + 0.25 V. But like I said, that's for hard switching.
If you want to go lower, delete the BC547 and use a higher hFE power transistor, like BD437.

[bdunham7]

Voltage at Rset is 0.75V, worst case Vbe of the pair roughly 1.72V. Vb of the pair would be 0.75V + 1.72V = 2.47V. Current needed on the base is 0.3A / 25 / 40 = 0.3 mA. So opamp should do the job.
Minimum Vce of the NPN pair I calculate as follows: Vbe(sat) BD139 + Vce(sat) BC847 = 1.25V in order circuit to work.

You can't discuss h_FE and V_sat at the same time, which is another way of saying the same thing as Benta.  You could make R18 smaller and try to have more base current, but I don't think you can have both transistors at or near saturation and simultaneously regulate the current, at least not in a stable manner.

[doktor pyta]

Isn't C-B junction of Q2 conducting when you apply low Vsupply?
You can measure voltage across R18 to check that.
So maybe use a logic level mosfet as a power transistor.

[moffy]

What you could do is measure the actual Vbe's rather than go off calculated as well as the output voltage of the opamp, and voltage across Rset and find out where the discrepancy lies.

[Rumpelstiltskin]

Thanks everyone for the replies!

...
If you want to go lower, delete the BC547 and use a higher hFE power transistor, like BD437.

For bd437 I see minimum is hFE is 85 and it can vary a lot of course. In this case it would introduce unacceptable error for load current. It needs to be set more precisely.

You can't discuss h_FE and V_sat at the same time, which is another way of saying the same thing as Benta.  You could make R18 smaller and try to have more base current, but I don't think you can have both transistors at or near saturation and simultaneously regulate the current, at least not in a stable manner.

I reduced R18 to 100 ohms and can't see the difference in regards of minimum Vce of the NPN pair.

What you could do is measure the actual Vbe's rather than go off calculated as well as the output voltage of the opamp, and voltage across Rset and find out where the discrepancy lies.

• Voltage across Rset is as expected 0.75V.
• Vbe of the BD139 is 0.8V and Vbe (BC847) is 0.87V.
• Measured base current across R18 is 7 mA roughly.
• Vout of the opamp is 3.15V (almost max).
• Circuit is working.

The whole point of the circuit was to reduce Pd across NPN by varying Vsupply and thus keeping Vce as low as possible.
I guess I'll just increase the minimum Vce requirement to 2V and attach a heatsink to the BD139.
Maybe there're still better ideas?

[moffy]

With those figures you have a total hfe of 42, which is around that for the BD139, is it possible the C and E of the BC847 are swapped? That would explain the low gain and lack of saturation of the BC847. If not it might be that the BC847 has no voltage connected to the collector, acting just as a diode.

[Vovk_Z]

0.3 A across a 2.5 Ohm shunt takes 0.75 VDC. That's possibly too much for a low-power or low-voltage current source.
Having precision RtoR input capable opamp you can use something about 0.050 .. 0.100 VDC voltage across the shunt (and put a respected voltage divider across an input of a circuit).
I use 0.1 VDC across a shunt even with general-use gelly-bean opamps like LM258/358.
So, I guess the shunt could be somewhere between 0.1 R ... 1.0 R. Or. Taking into account the stability it can be hard to work with a 0.1 R shunt but 0.20-0.22 R may work fine possibly. 
And better to change R19 from 10k to 1k - this can improve precision and stability.

[Rumpelstiltskin]

0.3 A across a 2.5 Ohm shunt takes 0.75 VDC. That's possibly too much for a low-power or low-voltage current source.
Having precision RtoR input capable opamp you can use something about 0.050 .. 0.100 VDC voltage across the shunt (and put a respected voltage divider across an input of a circuit).
I use 0.1 VDC across a shunt even with general-use gelly-bean opamps like LM258/358.
So, I guess the shunt could be somewhere between 0.1 R ... 1.0 R. Or. Taking into account the stability it can be hard to work with a 0.1 R shunt but 0.20-0.22 R may work fine possibly. 
And better to change R19 from 10k to 1k - this can improve precision and stability.

Thanks for advice. But in my case I monitor V_emitter with ADC and reducing active range would require additional parts to be installed. 2.5 ohms is a trade-off. Also reducing R19 won't introduce meaningful precision for me.

With those figures you have a total hfe of 42, which is around that for the BD139, is it possible the C and E of the BC847 are swapped? That would explain the low gain and lack of saturation of the BC847. If not it might be that the BC847 has no voltage connected to the collector, acting just as a diode.

Thanks a lot for help! Problem was in BD139, not in BC847. I don’t know how did a small transistor withstand 0.3 amps. I replaced big one and minimum Vce dropped to 0.75V.

[moffy]

Nicely done.